Leaflet platform for LibreOffice On-Line
========================================
This is the client part of LibreOffice On-Line. For the server part, see the
../loolwsd/README, and install it first.
Build dependencies
------------------
First you need to install 'jake'. As root, do:
npm install -g jake
npm install
npm is provided by the nodejs package.
Alternatively, you can use the provided zip (as a normal user):
unzip node_modules/modules.zip -d node_modules
A third way is to use npm as a user, but set its prefix to a directory where
you have write access. If you want that, you need to have an ~/.npmrc with the
line e.g.
prefix=/opt/npm
Building
--------
As a normal user:
jake build
cd plugins/draw-0.2.4
jake build
Running
-------
To see an example:
* run loolwsd, like:
./loolwsd --systemplate=${SYSTEMPLATE} --lotemplate=${MASTER}/instdir --childroot=${ROOTFORJAILS}
* open debug/document/document_simple_example.html in the browser, like:
firefox debug/document/document_simple_example.html
* modify the URL to allow loleaflet to connect to a server and open a document
http://web-page/document_simple_example.html?file_path=file:///PATH/TO/DOC&host=ws://localhost:9980
and you should see the document in the browser. In case anything goes wrong,
check the loolwsd console for the debugging output.
API & events
------------
Search:
- API:
map.search(text, [backward])
- events:
map.on('search', function (e) {}) (currently only fired when no search result is found) where:
+ e.originalPhrase = the phrase that has been searched for
+ e.count = number of results
Zoom:
- API:
map.zoomIn(amount)
map.zoomOut(amout)
map.getMinZoom()
map.getMaxZoom()
- events:
map.on('zoomend zoomlevelschange', function)
Edit, view, readOnly:
- API:
map.setPermission('edit' | 'view' | 'readonly')
- events:
map.on('updatepermission', function (e) {}) where:
+ e.perm == 'edit' | 'view' | 'readonly'
Buttons like Bold, Italic, Strike through etc.
- API:
map.toggleCommandState('Bold' | 'Italic' | 'Underline' | 'Strikeout' | 'AlighLeft', etc)
- events:
map.on('commandstatechanged', function (e) {}) where:
+ e.unoCmd == 'Bold' | 'Italic' etc.
+ e.state = 'true' | 'false'
Parts (like slides in presentation, or sheets in spreadsheets):
- API:
map.setPart('next' | 'prev' | partNumber)
map.getPartPreview(id, part, maxWidth, maxHeight, [options]) where:
+ id = the ID of the request so that the response can be identified
+ maxWidth / maxHeight are the desired dimensions of the preview, a smaller
image might be returned in order to keep the original ratio of the document
+ options = {autoUpdate: true} - automatically updates the previews
map.getNumberOfParts()
map.getCurrentPartNumber()
- events:
map.on('updateparts', function (e) {}) where:
+ e.currentPart is the current part
+ e.parts == the number of parts that the document has
+ e.docType == 'text' | 'spreadsheet' | 'presentation' | 'drawing' | 'other'
+ [e.partNames] if present, part names (e.g. sheet names)
map.on('invalidatepreview', function (e) {})
+ e.id = the preview's id
Statusindicator (when the document is loading):
- events
map.on('statusindicator', function (e) {}) where:
+ e.statusType = 'start' | 'setvalue' | 'finish' | 'loleafletloaded' | 'alltilesloaded'
+ e.value == a value from 0 to 100 indicating the status
if the statusType is 'setvalue
+ 'loleafletloaded' is fired when the JS code is initialized and the document
load request is sent and we're waiting for the tiles
+ 'alltilesloaded' is fired when all newly requested (empty tiles) have been loaded
it is not fired during pre-fetching and during editing
Save:
- API:
map.saveAs(url, [format, options])
Scroll (the following are measured in pixels):
- API:
+ options = An object with members: update (type: Boolean, default: false)
like {update: true}
map.scroll(x,y, options)
+ scroll right by 'x' and down by 'y' (or left and up if negative)
map.scrollDown(y, options)
+ scroll down by 'y' (or up if negative)
map.scrollRight(x, options)
+ scroll right by 'x' (or left if nevative)
map.scrollTop(y, options)
+ scroll to 'y' offset relative to the beginning of the document
map.scrollLeft(x, options)
+ scroll to 'x' offset relative to the beginning of the document
map.scrollOffset()
+ returns the scroll offset relative to the beginning of the document
map.getDocSize()
+ returns the document's size in pixels
map.getDocType()
+ returns 'text' | 'spreadsheet' | 'presentation' | 'drawing' | 'other'
- events
map.on('docsize', function (e) {}) where:
+ e.x = document width
+ e.y = document height
map.on('updatescrolloffset', function (e) {}) where:
+ e.x = difference between document's left and current view's left
(how much has the document been scrolled right)
+ e.y = difference between document's top and current view's top
(how much has the document been scrolled down)
- this event is fired when zooming and the current view is maintained but the
document shrinks or grow OR when the document is panned OR when the container is resized
map.on('scrollto', function (e) {}) where:
+ e.x = view's left position (so that the cursor/search result is in the center)
+ e.y = view's top position (so that the cursor/search result is in the center)
map.on('scrollby', function (e) {}) where:
+ e.x = the amount scrolled to the right (or left if negative)
+ e.y = the amount scrolled to the bottom (or top if negative)
Writer pages:
- API:
map.goToPage(page)
map.getNumberOfPages()
map.getCurrentPageNumber()
map.getDocPreview(id, maxWidth, maxHeight, x, y, width, height, [options])
+ id = the ID of the request so that the response can be identified
+ maxWidth / maxHeight are the desired dimensions of the preview, a smaller
image might be returned in order to keep the original ratio of the document
+ x/y = starting position, where to get the preview from
+ options = {autoUpdate: true} - automatically updates the previews
- events
map.on('pagenumberchanged', function (e) {}) where:
+ e.currentPage = the page on which the cursor lies
+ e.pages = number of pages
+ e.docType = document type, should be 'text'
map.on('invalidatepreview', function (e) {})
+ e.id = the preview's id
Styles:
- API:
map.getStyles()
+ returns a JSON object as a mapping of style families to a list of styles
map.setStyle(style, familyName)
- events:
map.on('updatestyles', function (e) {}) where:
e.styles = a JSON object as a mapping of style families to a list of styles
Error:
- events
map.on('error', function (e) {}) where
+ [e.msg] = a message describing the error
+ [e.cmd] = the command that caused the error
+ [e.kind] = the kind of error
Testing
-------
- to simulate an editing session and to get the tile loading times
+ open spec/tilebench.html in the browser
- to simulate a client opening several documents in the browser
+ open spec/loadtest.html in the browser
- to simulate a client opening several documents in the console
+ run: mocha headlessLoadTest.js
Contributing
------------
Code conventions:
* 'jake lint' should be run before commiting
* files should have unix line terminators (LF)
* tools to convert files: dos2unix or fromdos
Implementation details
----------------------
Loading a document:
The map should have the following options:
- server address
- doc - path to the document that will be loaded
- edit = the initial permission
- readOnly - whether the document is read only
- [timestamp] - optionally provided for remote documents
How zooming works:
The zoom level goes from 1 to 20 (those limits can be changed) and the initial
level is 10, which represents the 100% zoom level. The zoom factor is 1.2
Controls are added above the map in a div called "controls" is intended to be used as a toolbar.
There is no leaflet method of adding them in a separate div, so for now this is done in the html
document after the map initialization.
To enable scrollbars the map is placed above a div that contains a bigger div of
the document's size (a mock document). So the div under the map gets scrollbars which
are independent of the map's div, thus enabling us to link them to the map as needed.
When the user scrolls, the map is panned by the same amount as it would've been scrolled.
Also, some custom jquery scrollbars are used, to trigger the same scroll events across
browsers.